The decrease in overall incidence of TB in industrialised countries, together with the increasing concern for adverse events following BCG immunization, has led to important modifications of BCG policies in the last decades. This article adapts and validates - with surveillance data - a theoretical model estimating the impact of different national BCG vaccination policies on severe childhood TB in low to intermediate TB prevalent countries. The model shows that a universal BCG programme could be beneficial in settings with prevalence levels around 30 sputum smear positive per 100,000. In settings with prevalence levels below 15 per 100,000 the benefit of universal BCG vaccination should be carefully assessed, particularly where prevalence is below 5 per 100,000 and universal vaccination might lead to an excess of adverse events per case prevented. To this purpose the model also provides a tool to assess the theoretical impact of a policy change towards selective "high-risk groups" vaccination.